Archive
Weekly QuEST Discussion Topics, 30 July
QuEST 30 July 2021
We want to continue our previous discussions on issues with consciousness associated with the work of Ben Libet that suggest comments like:
‘What is done by what is called myself is, I feel, done by something greater than
myself in me’ – James Clerk Maxwell on his deathbed, 1879 (user illusion)
(implying ‘myself’ being the conscious agent)
• Reaction time experiment –
• Subjects demonstrated .25 secs – normal
• When he asked the subjects to increase their reaction time – NONE OF THEM COULD – when they tried to slow down slightly – more than a quarter of a second it leaped to .5 sec
• Conclusion – humans can react quickly but they cannot voluntarily react a little more slowly – if they react a little more slowly than they do instinctively they have to react consciously and that takes a lot longer.
• It may take ½ sec of activity in the sensory cortex before consciousness occurs but the subjective experience is assigned to an earlier point in time – namely the point in time when the stimulation occurred (I would suggest the moment in time when the faster representation update of the stimulus got posted – what we’ve called the Libet part of the representation has to be synchronized) synch’d with evoked potential
Consciousness lags behind but our subjective representation doesn’t – some have suggested a ‘veto theory’ – that is consciousness is to decide whether to execute decisions made subconsciously – this discussion is attempting to answer the challenge of Ancient Mike on the purpose of consciousness:
• The Scientific American one would be a good one to start. The Cell paper is a technical showing of how these questions are being asked in the field today (modeling, neuroscience, etc.). It doesn’t seem at all to be settled. Establishing biomarkers (brain measures) that index something like a “Free Won’t Mode,” (such as the Readiness Potential) for conscious controlled /unconscious automatic processing could work and have a lot of applications, and then through modeling experiments provide principles for Safe AI. Happy to dig deeper with you guys if it’s interesting. Cheers, Kevin
• https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(21)00093-0
We may also want to look at the details of the DARPA AGENT scenarios – and their applicability to our search for challenge problems that would enable the measurement of advances towards conscious computers.
We may also want to work towards some discussions on the latest article from Bengio, LeCun, Hinton:
https://cacm.acm.org/magazines/2021/7/253464-deep-learning-for-ai/fulltext
Turing lecture
Weekly QuEST Discussion Topics, 23 July
QuEST 23 July 2021
We will start this week with the BYOQ – Bring your own question? The idea is if anyone has spent time this week pondering ‘consciousness’ and has run up against an idea they want to discuss or ask the group’s opinion of this is the time to bring it up. You can submit the question early and we can post it or just bring it to the meeting.
After any new BYOQ issues are dealt with we will see if there are lingering thoughts from last week’s discussion. As a reminder last week focused on:
“If you have some free time, could you explain to me again the argument you made about consciousness today?”
• How does consciousness come into being, what is it physically, and what was it before it was “conscious?
• Perhaps more importantly, what isn’t consciousness? Plants? amoebas? (Dirt?).
• We can also talk about it at a quest meeting if you prefer. (It just that you seemed to make sense for a moment 🙂
We then want to continue our discussion hitting some high points on the Conductor Theory of Consciousness from our Intel colleague (Joscha Bach) – with some summary thoughts on how to develop challenge problems that focus on flexible AI solutions – also including our previous discussions on issues with consciousness associated with the work of Ben Libet that suggest comments like:
‘What is done by what is called myself is, I feel, done by something greater than
myself in me’ – James Clerk Maxwell on his deathbed, 1879 (user illusion)
(implying ‘myself’ being the conscious agent)
• Reaction time experiment –
• Subjects demonstrated .25 secs – normal
• When he asked the subjects to increase their reaction time – NONE OF THEM COULD – when they tried to slow down slightly – more than a quarter of a second it leaped to .5 sec
• Conclusion – humans can react quickly but they cannot voluntarily react a little more slowly – if they react a little more slowly than they do instinctively they have to react consciously and that takes a lot longer.
• It may take ½ sec of activity in the sensory cortex before consciousness occurs but the subjective experience is assigned to an earlier point in time – namely the point in time when the stimulation occurred (I would suggest the moment in time when the faster representation update of the stimulus got posted – what we’ve called the Libet part of the representation has to be synchronized) synch’d with evoked potential
• Consciousness lags behind but our subjective representation doesn’t
As a reminder our previous Bach discussions were focused on:
The Cortical Conductor Theory: Towards Addressing Consciousness in AI Models
Specifically we want to look at consistency with our QuEST S3Q Theory:
• The formation of the model is driven largely by data compression, i.e. by optimizing for the data structure that allows the best predictions of future observations, based on past observations.
• This principle has for instance been described by Ray Solomonoff (1964): The best possible model that a computational agent can form about its environment is the shortest program among those that best predict an observation from past observations, for all observations and past observations.
…
Experience is the simulation produced by neocortex – ** the what it is like **
• The elements of this simulation do not necessarily correspond to actual objects in the universe: they are statistical regularities that our mind discovered in the patterns at its systemic interface.
• Our experience is not directed on the pattern generator that is the universe, but on the simulation produced in our neocortex.
• Thus, our minds cannot experience and operate in an “outer” reality, but in a dream that is constrained by the available sensory input and the context of previous input (Bach 2011).
• Human cognition does not stop at generative simulations, however.
• We can abstract our mental representations into a conceptual manifold (figure 3).
• Concepts can be thought of as an address space for our sensory-motor scripts, and they allow the interpolation between objects, as well as the manipulation and generation of previously unknown objects via inference.
• The conceptual manifold can be organized and manipulated using grammatical language, which allows the synchronization of concepts between speakers, even in the absence of corresponding sensory-motor scripts.
…
• Phenomenal consciousness may simply be understood as the most recent memory of what our prefrontal cortex attended to.
• Thus, conscious experience is not an experience of being in the world, or in an inner space, but a memory.
• It is the reconstruction of a dream generated more than fifty brain areas, reflected in the protocol of a single region.
By directing attention on its own protocol, the conductor can store and recreate a memory of its own experience of being conscious
…
We may also want to work towards some discussions on the latest article from Bengio, LeCun, Hinton:
https://cacm.acm.org/magazines/2021/7/253464-deep-learning-for-ai/fulltext
Turing lecture
Weekly QuEST Discussion Topics, 16 July
QuEST 16 July 2021
We will start this week with the BYOQ – Bring your own question? The idea is if anyone has spent time this week pondering ‘consciousness’ and has run up against an idea they want to discuss or ask the group’s opinion of this is the time to bring it up. You can submit the question early and we can post it or just bring it to the meeting.
One such email thread this week from Ancient Mike – to our QuEST team senior mathematician Jared (we will start with this week):
• Jared,
If you have some free time, could you explain to me again the argument you made about consciousness today?
• How does consciousness come into being, what is it physically, and what was it before it was “conscious?
• Perhaps more importantly, what isn’t consciousness? Plants? amoebas? (Dirt?).
• We can also talk about it at a quest meeting if you prefer. (It just that you seemed to make sense for a moment 🙂
Best,
Mike
We also want to continue our discussion on the Conductor Theory of Consciousness from our Intel colleague (Joscha Bach):
The Cortical Conductor Theory: Towards Addressing Consciousness in AI Models
Specifically we want to look at consistency with our QuEST S3Q Theory:
• The formation of the model is driven largely by data compression, i.e. by optimizing for the data structure that allows the best predictions of future observations, based on past observations.
• This principle has for instance been described by Ray Solomonoff (1964): The best possible model that a computational agent can form about its environment is the shortest program among those that best predict an observation from past observations, for all observations and past observations.
…
Experience is the simulation produced by neocortex – ** the what it is like **
• The elements of this simulation do not necessarily correspond to actual objects in the universe: they are statistical regularities that our mind discovered in the patterns at its systemic interface.
• Our experience is not directed on the pattern generator that is the universe, but on the simulation produced in our neocortex.
• Thus, our minds cannot experience and operate in an “outer” reality, but in a dream that is constrained by the available sensory input and the context of previous input (Bach 2011).
• Human cognition does not stop at generative simulations, however.
• We can abstract our mental representations into a conceptual manifold (figure 3).
• Concepts can be thought of as an address space for our sensory-motor scripts, and they allow the interpolation between objects, as well as the manipulation and generation of previously unknown objects via inference.
• The conceptual manifold can be organized and manipulated using grammatical language, which allows the synchronization of concepts between speakers, even in the absence of corresponding sensory-motor scripts.
…
• Phenomenal consciousness may simply be understood as the most recent memory of what our prefrontal cortex attended to.
• Thus, conscious experience is not an experience of being in the world, or in an inner space, but a memory.
• It is the reconstruction of a dream generated more than fifty brain areas, reflected in the protocol of a single region.
By directing attention on its own protocol, the conductor can store and recreate a memory of its own experience of being conscious
…
We may also want to work towards some discussions on the latest article from Bengio, LeCun, Hinton:
https://cacm.acm.org/magazines/2021/7/253464-deep-learning-for-ai/fulltext
Turing lecture
Weekly QuEST Discussion Topics, 9 July
QuEST 9 July 2021
We want this week to move onto a series of presentation by some of our QuEST colleagues that attempt to find a common QuEST compliant framework that merges our diverse interest in Neurobiology, Cognitive models, artificial intelligence and fundamental mathematics that generalize across all of these areas to explain consciousness – specifically our S3Q model.
We might also, when time allows begin a discussion of the ‘Conductor Theory of Consciousness’ by Joscha Bach. There are several online video presentations / interviews with Joscha Bach and also there is the article from BICA –
The Cortical Conductor Theory:
Towards Addressing Consciousness in AI Models
Joscha Bach (joscha@bach.ai)
Harvard Program for Evolutionary Dynamics, One Brattle Square #6
Cambridge, MA 02139 USA
AI models of the mind rarely discuss the so called “hard problem” of consciousness. Here, I will sketch informally possible functional explanation for phenomena consciousness: the conductor theory of consciousness (CTC).
Unlike IIT, CTC is a functionalist model of consciousness with similarity to other functionalist approaches, such as the ones suggested by Dennett and Graziano.
Quest Group's Blog 